Publications

Abstract

In the CLARIN infrastructure various national projects have started initiatives to allow users of the infrastructure to create chains or workflows of web services. The Component Metadata (CMD) core model for web services described in this paper tries to align the metadata descriptions of these various initiatives. This should allow chaining/workflow engines to find matching and invoke services. The paper describes the landscape of web services architectures and the state of the national initiatives. Based on this a CMD core model for CLARIN is proposed, which, within some limits, can be adapted to the specific needs of an initiative by the standard facilities of CMD. The paper closes with the current state and usage of the model and a look into the future.

Abstract

ISO Technical Committee 37, Terminology and other language and content resources, established an ISO 12620:2009 based Data Category Registry (DCR), called ISOcat (see http://www.isocat.org), to foster semantic interoperability of linguistic resources. However, this goal can only be met if the data categories are reused by a wide variety of linguistic resource types. A resource indicates its usage of data categories by linking to them. The small DC Reference XML vocabulary is used to embed links to data categories in XML documents. The link is established by an URI, which servers as the Persistent IDentifier (PID) of a data category. This paper discusses the efforts to mimic the same approach for RDF-based resources. It also introduces the RDF quad store based Relation Registry RELcat, which enables ontological relationships between data categories not supported by ISOcat and thus adds an extra level of linguistic knowledge.

Abstract

The ISOcat Data Category Registry contains basically a flat and easily extensible list of data category specifications. To foster reuse and standardization only very shallow relationships among data categories are stored in the registry. However, to assist crosswalks, possibly based on personal views, between various (application) domains and to overcome possible proliferation of data categories more types of ontological relationships need to be specified. RELcat is a first prototype of a Relation Registry, which allows storing arbitrary relationships. These relationships can reflect the personal view of one linguist or a larger community. The basis of the registry is a relation type taxonomy that can easily be extended. This allows on one hand to load existing sets of relations specified in, for example, an OWL (2) ontology or SKOS taxonomy. And on the other hand allows algorithms that query the registry to traverse the stored semantic network to remain ignorant of the original source vocabulary. This paper describes first experiences with RELcat and explains some initial design decisions.

Abstract

Automatic Language Identification of written texts is a well-established area of research in Computational Linguistics. State-of-the-art algorithms often rely on n-gram character models to identify the correct language of texts, with good results seen for European languages. In this paper we propose the use of a character n-gram model and a word n-gram language model for the automatic classification of two written varieties of Portuguese: European and Brazilian. Results reached 0.998 for accuracy using character 4-grams.

Abstract

Impaired emotion regulation may underlie exaggerated emotional reactivity in patients with obsessive compulsive disorder (OCD), yet instructed emotion regulation has never been studied in the disorder. METHOD: This study aimed to assess the neural correlates of emotion processing and regulation in 43 medication-free OCD patients and 38 matched healthy controls, and additionally test if these can be modulated by stimulatory (patients) and inhibitory (controls) repetitive transcranial magnetic stimulation (rTMS) over the left dorsolateral prefrontal cortex (dlPFC). Participants performed an emotion regulation task during functional magnetic resonance imaging before and after a single session of randomly assigned real or sham rTMS. Effect of group and rTMS were assessed on self-reported distress ratings and brain activity in frontal-limbic regions of interest. RESULTS: Patients had higher distress ratings than controls during emotion provocation, but similar rates of distress reduction after voluntary emotion regulation. OCD patients compared with controls showed altered amygdala responsiveness during symptom provocation and diminished left dlPFC activity and frontal-amygdala connectivity during emotion regulation. Real v. sham dlPFC stimulation differentially modulated frontal-amygdala connectivity during emotion regulation in OCD patients. CONCLUSIONS: We propose that the increased emotional reactivity in OCD may be due to a deficit in emotion regulation caused by a failure of cognitive control exerted by the dorsal frontal cortex. Modulatory rTMS over the left dlPFC may influence automatic emotion regulation capabilities by influencing frontal-limbic connectivity.

Abstract

Arbil is an application designed to create and manage metadata for research data and to arrange this data into a structure appropriate for archiving. The metadata is displayed in tables, which allows an overview of the metadata and the ability to populate and update many metadata sections in bulk. Both IMDI and Clarin metadata formats are supported and Arbil has been designed as a local application so that it can also be used offline, for instance in remote field sites. The metadata can be entered in any order or at any stage that the user is able; once the metadata and its data are ready for archiving and an Internet connection is available it can be exported from Arbil and in the case of IMDI it can then be transferred to the main archive via LAMUS (archive management and upload system).

Abstract

In three cross-modal priming experiments we asked whether adaptation to a foreign-accented speaker is automatic, and whether adaptation can be seen after a long delay between initial exposure and test. Dutch listeners were exposed to a Hebrew-accented Dutch speaker with two types of Dutch words: those that contained [ɪ] (globally accented words), and those in which the Dutch [i] was shortened to [ɪ] (specific accent marker words). Experiment 1, which served as a baseline, showed that native Dutch participants showed facilitatory priming for globally accented, but not specific accent, words. In experiment 2, participants performed a 3.5-minute phoneme monitoring task, and were tested on their comprehension of the accented speaker 24 hours later using the same cross-modal priming task as in experiment 1. During the phoneme monitoring task, listeners were asked to detect a consonant that was not strongly accented. In experiment 3, the delay between exposure and test was extended to 1 week. Listeners in experiments 2 and 3 showed facilitatory priming for both globally accented and specific accent marker words. Together, these results show that adaptation to a foreign-accented speaker can be rapid and automatic, and can be observed after a prolonged delay in testing.

Abstract

Introduction Speech research saw a dramatic change in paradigm in the 90-ies. While earlier the discussion was dominated by a phoneticians’ approach who knew about phenomena in the speech signal, the situation completely changed after stochastic machinery such as Hidden Markov Models [1] and Artificial Neural Networks [2] had been introduced. Speech processing was now dominated by a purely mathematic approach that basically ignored all existing knowledge about the speech production process and the perception mechanisms. The key was now to construct a large enough training set that would allow identifying the many free parameters of such stochastic engines. In case that the training set is representative and the annotations of the training sets are widely ‘correct’ we could assume to get a satisfyingly functioning recognizer. While the success of knowledge-based systems such as Hearsay II [3] was limited, the statistically based approach led to great improvements in recognition rates and to industrial applications.

Abstract

5.1 Einführung in den Forschungsbereich Die Psycholinguistik ist der Bereich der Linguistik, der sich mit dem Zusammenhang zwischen menschlicher Sprache und dem Denken und anderen mentalen Prozessen beschäftigt, d.h. sie stellt sich einer Reihe von essentiellen Fragen wie etwa (1) Wie schafft es unser Gehirn, im Wesentlichen akustische und visuelle kommunikative Informationen zu verstehen und in mentale Repräsentationen umzusetzen? (2) Wie kann unser Gehirn einen komplexen Sachverhalt, den wir anderen übermitteln wollen, in eine von anderen verarbeitbare Sequenz von verbalen und nonverbalen Aktionen umsetzen? (3) Wie gelingt es uns, in den verschiedenen Phasen des Lebens Sprachen zu erlernen? (4) Sind die kognitiven Prozesse der Sprachverarbeitung universell, obwohl die Sprachsysteme derart unterschiedlich sind, dass sich in den Strukturen kaum Universalien finden lassen?

Abstract

According to a widely-held view among various scholars, olfaction is inferior to other human senses. It is also believed by many that languages do not have words for describing smells. Data collected among the Maniq, a small population of nomadic foragers in southern Thailand, challenge the above claims and point to a great linguistic and cultural elaboration of odor. This article presents evidence of the importance of olfaction in indigenous rituals and beliefs, as well as in the lexicon. The results demonstrate the richness and complexity of the domain of smell in Maniq society and thereby challenge the universal paucity of olfactory terms and insignificance of olfaction for humans.

Abstract

To identify genetic variants associated with refractive astigmatism in the general population, meta-analyses of genome-wide association studies were performed for: White Europeans aged at least 25 years (20 cohorts, N = 31,968); Asian subjects aged at least 25 years (7 cohorts, N = 9,295); White Europeans aged <25 years (4 cohorts, N = 5,640); and all independent individuals from the above three samples combined with a sample of Chinese subjects aged <25 years (N = 45,931). Participants were classified as cases with refractive astigmatism if the average cylinder power in their two eyes was at least 1.00 diopter and as controls otherwise. Genome-wide association analysis was carried out for each cohort separately using logistic regression. Meta-analysis was conducted using a fixed effects model. In the older European group the most strongly associated marker was downstream of the neurexin-1 (NRXN1) gene (rs1401327, P = 3.92E−8). No other region reached genome-wide significance, and association signals were lower for the younger European group and Asian group. In the meta-analysis of all cohorts, no marker reached genome-wide significance: The most strongly associated regions were, NRXN1 (rs1401327, P = 2.93E−07), TOX (rs7823467, P = 3.47E−07) and LINC00340 (rs12212674, P = 1.49E−06). For 34 markers identified in prior GWAS for spherical equivalent refractive error, the beta coefficients for genotype versus spherical equivalent, and genotype versus refractive astigmatism, were highly correlated (r = −0.59, P = 2.10E−04). This work revealed no consistent or strong genetic signals for refractive astigmatism; however, the TOX gene region previously identified in GWAS for spherical equivalent refractive error was the second most strongly associated region. Analysis of additional markers provided evidence supporting widespread genetic co-susceptibility for spherical and astigmatic refractive errors.

Abstract

In het onderwijs wordt aangenomen dat hardop en stillezen dezelfde processen zijn. In dit onderzoek wordt gekeken naar het verschil tussen hardop en stillezen wat betreft leessnelheid en tekstbegrip bij 90 kinderen uit groep 4. Ook wordt de invloed van de cognitieve vaardigheden fonologisch bewustzijn, benoemsnelheid en visuele aandachtsspanne op de verschillende leesmodi onderzocht. De participanten lazen stil sneller, maar begrepen de tekst beter hardop. De cognitieve vaardigheden correleerden met hardop en stillezen wat betreft leessnelheid, maar hingen in beide leesmodi niet samen met tekstbegrip. Hoewel hardop en stillezen samenhangen, onderstrepen deze bevindingen dat het verschillende leesmodi zijn.

Abstract

Diffusion tensor imaging (DTI) and a longitudinal language learning approach were applied to investigate the relationship between the achieved second language (L2) proficiency during L2 learning and the reorganization of structural connectivity between core language areas. Language proficiency tests and DTI scans were obtained from German students before and after they completed an intensive 6-week course of the Dutch language. In the initial learning stage, with increasing L2 proficiency, the hemispheric dominance of the BA6-temporal pathway (mainly along the arcuate fasciculus) shifted from the left to the right hemisphere. With further increased proficiency, however, lateralization dominance was again found in the left BA6-temporal pathway. This result is consistent with reports in the literature that imply a stronger involvement of the right hemisphere in L2-processing especially for less proficient L2-speakers. This is the first time that a L2-proficiency-dependent laterality shift in structural connectivity of language pathways during L2 acquisition has been observed to shift from left to right, and back to left hemisphere dominance with increasing L2-proficiency. We additionally find that changes in fractional anisotropy values after the course are related to the time elapsed between the two scans. The results suggest that structural connectivity in (at least part of) the perisylvian language network may be subject to fast dynamic changes following language learning

Abstract

We carried out the first study on the relationship between individual language aptitude and structural connectivity of language pathways in the adult brain. We measured four components of language aptitude (vocabulary learning, VocL; sound recognition, SndRec; sound-symbol correspondence, SndSym; and grammatical inferencing, GrInf) using the LLAMA language aptitude test (Meara, 2005). Spatial working memory (SWM), verbal working memory (VWM) and IQ were also measured as control factors. Diffusion Tensor Imaging (DTI) was employed to investigate the structural connectivity of language pathways in the perisylvian language network. Principal Component Analysis (PCA) on behavioural measures suggests that a general ability might be important to the first stages of L2 acquisition. It also suggested that VocL, SndSy and SWM are more closely related to general IQ than SndRec and VocL, and distinguished the tasks specifically designed to tap into L2 acquisition (VocL, SndRec,SndSym and GrInf) from more generic measures (IQ, SWM and VWM). Regression analysis suggested significant correlations between most of these behavioural measures and the structural connectivity of certain language pathways, i.e., VocL and BA47-Parietal pathway, SndSym and inter-hemispheric BA45 pathway, GrInf and BA45-Temporal pathway and BA6-Temporal pathway, IQ and BA44-Parietal pathway, BA47-Parietal pathway, BA47-Temporal pathway and inter-hemispheric BA45 pathway, SWM and inter-hemispheric BA6 pathway and BA47-Parietal pathway, and VWM and BA47-Temporal pathway. These results are discussed in relation to relevant findings in the literature.

Abstract

In recent decades, neuroimaging studies on the neural infrastructure of language are usually (or mostly) conducted with certain on-line language processing tasks. These functional neuroimaging studies helped to localize the language areas in the brain and to investigate the brain activity during explicit language processing. However, little is known about what is going on with the language areas when the brain is ‘at rest’, i.e., when there is no explicit language processing running. Taking advantage of the fcMRI and DTI techniques, this thesis is able to investigate the language function ‘off-line’ at the neuronal network level and the connectivity among language areas in the brain. Based on patient studies, the traditional, classical model on the perisylvian language network specifies a “Broca’ area – Arcuate Fasciculus – Werinicke’s area” loop (Ojemann 1991). With the help of modern neuroimaging techniques, researchers have been able to track language pathways that involve more brain structures than are in the classical model, and relate them to certain language functions. In such a background, a large part of this thesis made a contribution to the study of the topology of the language networks. It revealed that the language networks form a topographical functional connectivity pattern in the left hemisphere for the right-handers. This thesis also revealed the importance of structural hubs, such as Broca’s and Wernicke’s areas, which have more connectivity to other brain areas and play a central role in the language networks. Furthermore, this thesis revealed both functionally and structurally lateralized language networks in the brain. The consistency between what is found in this thesis and what has been known from previous functional studies seems to suggest, that the human brain is optimized and ‘ready’ for the language function even when there is currently no explicit language-processing running.

Abstract

Four experiments investigated the role of the syllable in Chinese spoken word production. Chen, Chen and Ferrand (2003) reported a syllable priming effect when primes and targets shared the first syllable using a masked priming paradigm in Chinese. Our Experiment 1 was a direct replication of Chen et al.'s (2003) Experiment 3 employing CV (e. g., /ba2.ying2/, strike camp) and CVG (e. g., /bai2.shou3/, white haired) syllable types. Experiment 2 tested the syllable priming effect using different syllable types: e. g., CV (/qi4.qiu2/, balloon) and CVN (/qing1.ting2/, dragonfly). Experiment 3 investigated this issue further using line drawings of common objects as targets that were preceded either by a CV (e. g., /qi3/, attempt), or a CVN (e. g., /qing2/, affection) prime. Experiment 4 further examined the priming effect by a comparison between CV or CVN priming and an unrelated priming condition using CV-NX (e. g., /mi2.ni3/, mini) and CVN-CX (e. g., /min2.ju1/, dwellings) as target words. These four experiments consistently found that CV targets were named faster when preceded by CV primes than when they were preceded by CVG, CVN or unrelated primes, whereas CVG or CVN targets showed the reverse pattern. These results indicate that the priming effect critically depends on the match between the structure of the prime and that of the first syllable of the target. The effect obtained in this study was consistent across different stimuli and different tasks (word and picture naming), and provides more conclusive and consistent data regarding the role of the syllable in Chinese speech production.

Abstract

Automatic Language Identification of written texts is a well-established area of research in Computational Linguistics. State-of-the-art algorithms often rely on n-gram character models to identify the correct language of texts, with good results seen for European languages. In this paper we propose the use of a character n-gram model and a word n-gram language model for the automatic classification of two written varieties of Portuguese: European and Brazilian. Results reached 0.998 for accuracy using character 4-grams.

Abstract

This study presents a new language identification model for pluricentric languages that uses n-gram language models at the character and word level. The model is evaluated in two steps. The first step consists of the identification of two varieties of Spanish (Argentina and Spain) and two varieties of French (Quebec and France) evaluated independently in binary classification schemes. The second step integrates these language models in a six-class classification with two Portuguese varieties.

Abstract

This article reports on results from a broad crosslinguistic study based on data from thirty-five signed languages around the world. The study is the first of its kind, and the typological generalizations presented here cover the domain of interrogative structures as they appear across a wide range of geographically and genetically distinct signed languages. Manual and nonmanual ways of marking basic types of questions in signed languages are investigated. As a result, it becomes clear that the range of crosslinguistic variation is extensive for some subparameters, such as the structure of question-word paradigms, while other parameters, such as the use of nonmanual expressions in questions, show more similarities across signed languages. Finally, it is instructive to compare the findings from signed language typology to relevant data from spoken languages at a more abstract, crossmodality level.

Abstract

This article presents a typology of negative constructions across a substantial number of sign languages from around the globe. After situating the topic within the wider context of linguistic typology, the main negation strategies found across sign languages are described. Nonmanual negation includes the use of head movements and facial expressions for negation and is of great importance in sign languages as well as particularly interesting from a typological point of view. As far as manual signs are concerned, independent negative particles represent the dominant strategy, but there are also instances of irregular negation in most sign languages. Irregular negatives may take the form of suppletion, cliticisation, affixing, or internal modification of a sign. The results of the study lead to interesting generalisations about similarities and differences between negatives in signed and spoken languages.

Abstract

The book is a unique collection of research on sign languages that have emerged in rural communities with a high incidence of, often hereditary, deafness. These sign languages represent the latest addition to the comparative investigation of languages in the gestural modality, and the book is the first compilation of a substantial number of different "village sign languages".Written by leading experts in the field, the volume uniquely combines anthropological and linguistic insights, looking at both the social dynamics and the linguistic structures in these village communities. The book includes primary data from eleven different signing communities across the world, including results from Jamaica, India, Turkey, Thailand, and Bali. All known village sign languages are endangered, usually because of pressure from larger urban sign languages, and some have died out already. Ironically, it is often the success of the larger sign language communities in urban centres, their recognition and subsequent spread, which leads to the endangerment of these small minority sign languages. The book addresses this specific type of language endangerment, documentation strategies, and other ethical issues pertaining to these sign languages on the basis of first-hand experiences by Deaf fieldworkers

Abstract

In the word-learning domain, both adults and young children are able to find the correct referent of a word from highly ambiguous contexts that involve many words and objects by computing distributional statistics across the co-occurrences of words and referents at multiple naming moments (Yu & Smith, 2007; Smith & Yu, 2008). However, there is still debate regarding how learners accumulate distributional information to learn object labels in natural learning environments, and what underlying learning mechanism learners are most likely to adopt. Using the Human Simulation Paradigm (Gillette, Gleitman, Gleitman & Lederer, 1999), we found that participants’ learning performance gradually improved and that their ability to remember and carry over partial knowledge from past learning instances facilitated subsequent learning. These results support the statistical learning model that word learning is a continuous process.

Abstract

Osteoblast induction and differentiation in developing long bones is dynamically controlled by the opposing action of transcriptional activators and repressors. In contrast to the long list of activators that have been discovered over past decades, the network of repressors is not well-defined. Here we identify the expression of Foxp1/2/4 proteins, comprised of Forkhead-box (Fox) transcription factors of the Foxp subfamily, in both perichondrial skeletal progenitors and proliferating chondrocytes during endochondral ossification. Mice carrying loss-of-function and gain-of-function Foxp mutations had gross defects in appendicular skeleton formation. At the cellular level, over-expression of Foxp1/2/4 in chondroctyes abrogated osteoblast formation and chondrocyte hypertrophy. Conversely, single or compound deficiency of Foxp1/2/4 in skeletal progenitors or chondrocytes resulted in premature osteoblast differentiation in the perichondrium, coupled with impaired proliferation, survival, and hypertrophy of chondrocytes in the growth plate. Foxp1/2/4 and Runx2 proteins interacted in vitro and in vivo, and Foxp1/2/4 repressed Runx2 transactivation function in heterologous cells. This study establishes Foxp1/2/4 proteins as coordinators of osteogenesis and chondrocyte hypertrophy in developing long bones and suggests that a novel transcriptional repressor network involving Foxp1/2/4 may regulate Runx2 during endochondral ossification.

Abstract

Face-selective regions (FSRs) are among the most widely studied functional regions in the human brain. However, individual variability of the FSRs has not been well quantified. Here we use functional magnetic resonance imaging (fMRI) to localize the FSRs and quantify their spatial and functional variabilities in 202 healthy adults. The occipital face area (OFA), posterior and anterior fusiform face areas (pFFA and aFFA), posterior continuation of the superior temporal sulcus (pcSTS), and posterior and anterior STS (pSTS and aSTS) were delineated for each individual with a semi-automated procedure. A probabilistic atlas was constructed to characterize their interindividual variability, revealing that the FSRs were highly variable in location and extent across subjects. The variability of FSRs was further quantified on both functional (i.e., face selectivity) and spatial (i.e., volume, location of peak activation, and anatomical location) features. Considerable interindividual variability and rightward asymmetry were found in all FSRs on these features. Taken together, our work presents the first effort to characterize comprehensively the variability of FSRs in a large sample of healthy subjects, and invites future work on the origin of the variability and its relation to individual differences in behavioral performance. Moreover, the probabilistic functional atlas will provide an adequate spatial reference for mapping the face network.

Abstract

Semantic unification, the process by which small blocks of semantic information are combined into a coherent utterance, has been studied with various types of tasks. However, whether the brain activations reported in these studies are attributed to semantic unification per se or to other task-induced concomitant processes still remains unclear. The neural basis for semantic unification in sentence comprehension was examined using event-related potentials (ERP) and functional Magnetic Resonance Imaging (fMRI). The semantic unification load was manipulated by varying the goodness of fit between a critical word and its preceding context (in high cloze, low cloze and violation sentences). The sentences were presented in a serial visual presentation mode. The participants were asked to perform one of three tasks: semantic congruency judgment (SEM), silent reading for comprehension (READ), or font size judgment (FONT), in separate sessions. The ERP results showed a similar N400 amplitude modulation by the semantic unification load across all of the three tasks. The brain activations associated with the semantic unification load were found in the anterior left inferior frontal gyrus (aLIFG) in the FONT task and in a widespread set of regions in the other two tasks. These results suggest that the aLIFG activation reflects a semantic unification, which is different from other brain activations that may reflect task-specific strategic processing.

Abstract

Neural correlates of lexical stress were studied using the mismatch negativity (MMN) component in event-related potentials. The MMN responses were expected to reveal the encoding of stress information into long-term memory and the contributions of prosodic features such as fundamental frequency (F0) and intensity toward lexical access. In a passive oddball paradigm, neural responses to changes in F0, intensity, and in both features together were recorded for words and pseudowords. The findings showed significant differences not only between words and pseudowords but also between prosodic features. Early processing of prosodic information in words was indexed by an intensity-related MMN and an F0-related P200. These effects were stable at right-anterior and mid-anterior regions. At a later latency, MMN responses were recorded for both words and pseudowords at the mid-anterior and posterior regions. The P200 effect observed for F0 at the early latency for words developed into an MMN response. Intensity elicited smaller MMN for pseudowords than for words. Moreover, a larger brain area was recruited for the processing of words than for the processing of pseudowords. These findings suggest earlier and higher sensitivity to prosodic changes in words than in pseudowords, reflecting a language-related process. The present study, therefore, not only establishes neural correlates of lexical stress but also confirms the presence of long-term memory traces for prosodic information in the brain.

Abstract

How does language comprehension interact with motor activity? We investigated the conditions under which comprehending an action sentence affects people's balance. We performed two experiments to assess whether sentences describing forward or backward movement modulate the lateral movements made by subjects who made sensibility judgments about the sentences. In one experiment subjects were standing on a balance board and in the other they were seated on a balance board that was mounted on a chair. This allowed us to investigate whether the action compatibility effect (ACE) is robust and persists in the face of salient incompatibilities between sentence content and subject movement. Growth-curve analysis of the movement trajectories produced by the subjects in response to the sentences suggests that the ACE is indeed robust. Sentence content influenced movement trajectory despite salient inconsistencies between implied and actual movement. These results are interpreted in the context of the current discussion of embodied, or grounded, language comprehension and meaning representation.

Abstract

Classifiers (currently also called 'depicting handshapes'), are observed in almost all signed languages studied to date and form a well-researched topic in sign language linguistics. Yet, these elements are still subject to much debate with respect to a variety of matters. Several different categories of classifiers have been posited on the basis of their semantics and the linguistic context in which they occur. The function(s) of classifiers are not fully clear yet. Similarly, there are differing opinions regarding their structure and the structure of the signs in which they appear. Partly as a result of comparison to classifiers in spoken languages, the term 'classifier' itself is under debate. In contrast to these disagreements, most studies on the acquisition of classifier constructions seem to consent that these are difficult to master for Deaf children. This article presents and discusses all these issues from the viewpoint that classifiers are linguistic elements.

Abstract

This paper explores the expression of multiple entities in Turkish Sign Language (Türk İşaret Dili; TİD), a less well-studied sign language. It aims to provide a comprehensive description of the ways and frequencies in which entity plurality in this language is expressed, both within and outside the noun phrase. We used a corpus that includes both elicited and spontaneous data from native signers. The results reveal that most of the expressions of multiple entities in TİD are iconic, spatial strategies (i.e. localization and spatial plural predicate inflection) none of which, we argue, should be considered as genuine plural marking devices with the main aim of expressing plurality. Instead, the observed devices for localization and predicate inflection allow for a plural interpretation when multiple locations in space are used. Our data do not provide evidence that TİD employs (productive) morphological plural marking (i.e. reduplication) on nouns, in contrast to some other sign languages and many spoken languages. We relate our findings to expression of multiple entities in other signed languages and in spoken languages and discuss these findings in terms of modality effects on expression of multiple entities in human language.